SIParCS 2016 - Pulong Ma

Pulong Ma, University of Cincinnati

DAbayes: Climate-Change Detection and Attribution Based on the Bayesian Hierarchical Model

(Slides)  (Recorded Talk)

Climate change is happening, and human forcing is suspected to be one of the causes. By considering climate-change detection and attribution as a multivariate spatial or spatio-temporal regression problem, we are able to understand whether and how the climate has changed due to anthropogenic and natural forcing scenarios. A Bayesian hierarchical statistical model has been developed for the detection and attribution problem, and we created an R package called "DAbayes" to implement the statistical model in R, which allows sequential and parallel Bayesian inference (e.g., Markov Chain Monte Carlo algorithm) via Bayesian model averaging. The DAbayes package incorporates both a simulation example and real data analysis. For the latter, observed or reconstructed measurements (e.g., temperature change) and general circulation model (GCM) outputs under several forcing scenarios are used to illustrate the statistical problem.

Mentors: Dorit Hammerling, Doug Nychka, Raghu Prasanna Kumar, Rory Kelly, CISL